Ideas and perspectives: Using meta-omics to unravel biogeochemical changes from cell to planetary scales

Increased human impacts on Earth systems are radically altering biogeochemical cycles. While long-term environmental observatories and Earth System Models (ESMs) provide valuable insights into the mechanisms of nutrient dynamics, their performance is limited at the fine spatial scales controlled by the functional diversity of plant and microbial communities. This gap in our understanding concerning the roles of microbial diversity and plant-microbial interactions in decomposition and nutrient dynamics extends across many global ecosystems. Recent advances in meta-omics technologies, including metagenomics, metatranscriptomics, metaproteomics, and metabolomics, offer a wide array of tools for assessing metabolic to genetic to evolutionary drivers of ecosystem functioning. Here, we explore the integration of meta-omics with traditional ecological approaches to examine responses to global environmental changes. We present case studies from diverse environments – soils, aquatic systems, clouds, and paleoarchives – demonstrating how meta-omics can unravel the roles of microbial diversity, metabolic pathways, and trait distributions critical to understanding greenhouse gas fluxes, nutrient cycling, and biogeochemistry. Although meta-omics is still beset with challenges including data heterogeneity arising from wide-ranging methods, omics-derived traits, kinetic parameters, and machine learning tools can be used to enhance ESM predictive capability. For example, emerging applications of meta-omics to ancient environmental DNA are extending our capacity to link historical patterns with future projections, offering a long-term perspective on ecosystem dynamics. This review highlights the potential of integrating omics with experimental manipulations alongside existing monitoring and modelling efforts to refine predictions of ecosystem responses to natural and anthropogenic-driven environmental changes. Because omics approaches cross a range of scientific domains, they could be used to foster collaboration and even integration within existing models, thus laying the foundation for informed conservation and ecosystem management strategies from local to global scales.

1 Introduction

Over the last century, the over-exploitation of natural resources through farming, mining, construction, and other industrial practices has intensified pressure on the biosphere, hydrosphere, and atmosphere (Rockström et al., 2023). The resulting losses of natural habitats have driven rapid shifts in the biogeochemical properties of ecosystems (Lewis and Maslin, 2015; Waters et al., 2016). These human-induced changes in global biogeochemical cycles alter energy fluxes that contribute to our current climate and biodiversity crises. The resulting pollution and environmental degradation that have been measured and communicated for decades by geoscientists and biologists necessitate continued biogeoscience research internationally (reports of the Intergovernmental Panel on Climate Change).

To understand the impacts on biogeochemical processes, long-term observatory studies with large-scale ground-based or ocean-based infrastructures – such as LTER (terrestrial and aquatic systems across the USA), NEON (diverse U.S. ecosystems), ICOS (European forests, wetlands, and grasslands), ITEX (Arctic tundra), BATS and SPOT (Atlantic and Pacific oceanic waters), and HOT (subtropical North Pacific Ocean) – were established to document the dynamic responses of ecosystems to changes in climate, species invasions, and management impacts (Franz et al., 2018; Trowbridge et al., 2019; Thibault et al., 2023). These local scale observatories are often equipped with instruments to measure the water, energy and greenhouse gas (GHGs) fluxes (CO₂, CH₄, N₂O) at the interface between surface ecosystems (soil, plant canopies, water bodies) and the atmosphere, providing valuable insights on the response of ecosystem processes, that underpin GHG budgets, to short (extreme and seasonal weather systems) and long term changes in climate and management. When these local scale measurements are coordinated across networks, the spatial and temporal variations in GHG budgets and ecological processes can be assessed over larger scales and used to generate predictive models, which in turn can be confronted at large scale by satellite and atmospheric station observations.

In addition to modern-day monitoring, the information preserved in sedimentary archives provides knowledge about past environmental changes and their potential consequences on biodiversity (Gregory-Eaves and Smol, 2024). The use of organic climate proxies (contained in tree rings, pollen, charcoal, corals, diatoms and foraminifera) and inorganic proxies (contained in varves, stalagmites, sediments and ice cores) provide a direct link to the effects of environmental change on ecosystem properties, such as changes in functional diversity, species abundances, and biogeographic distributions. For example, by identifying traits linked to present-day climate niches, researchers can extrapolate to past and future scenarios, using geological climate parallels to anticipate ecosystem responses (McElwain et al., 2024).

To date, most large-scale and long-term infrastructures have shown that representing a certain level of plant functional diversity can improve the predictive accuracy of Earth System Models (ESMs), particularly in simulating interactions among carbon, water, and nutrient cycles (Wullschleger et al., 2014; Anderegg et al., 2022). However, ESMs still face considerable limitations in capturing the full complexity of biogeochemical processes and energy fluxes, especially those mediated by microbial communities. These models often underrepresent key microbial functions that influence decomposition, nutrient cycling, and GHG fluxes. While primary production by higher plants is relatively well parameterized in many land surface models, the roles of microbial primary producers, heterotrophic decomposers, and organisms involved in secondary metabolism remain poorly understood and largely absent from ESM formulations. This gap includes processes such as microbial biomass growth, litter degradation, and the regulation of carbon use efficiency, which all influence the fate of organic matter and the balance between carbon storage and release (He et al., 2024b, 2026). The ecophysiology of these microbial communities has a direct feedback on the chemistry of the atmosphere and consequently climate (Lovelock and Margulis, 1974; Conrad, 1994; Monson and Holland, 2001). Yet, integrating mechanistic representations of microbial abundance, diversity, and function into ESMs remains incomplete, due to uncertainty about which processes are most critical, how to describe them mathematically, and how to scale them appropriately across spatial and temporal dimensions.

Meta-omics data, which encompass the comprehensive analyses of genes, transcripts, proteins, and metabolites of communities, provide powerful insights to bridge this knowledge gap. For example, molecular ecology techniques (e.g., metabarcoding, metagenomics, metatranscriptomics, metaproteomics, and metabolomics) used to study functional microbial ecology provide new ways to quantify and identify GHG production hotspots with novel metrics based on key traits (Frostegård et al., 2022). If implemented across networks that monitor large-scale and long-term changes in ecology and function (e.g., flux measurement sites) or through paleoecological archives, an omics framework could provide an important bridge between disciplines (Fig. 1). For example, microbial ecologists could then help identify “keystone” organisms that disproportionately impact ecosystem functioning, such as regulating GHG flux rates or those responding to climate variations.

Figure 1 Conceptual framework for advancing our understanding and prediction of Earth's biogeochemical cycles under environmental change, by integrating omics data into long-term monitoring and modelling efforts. Red: Earth environmental spheres and ecosystems; blue: microbial cells and microbiomes interacting with biogeochemical processes critical in global biogeochemical cycles; yellow: long-term monitoring actions that supply key knowledge and response variables for biogeochemical models; green: integration of different levels of functional omics data to complement the environmental datasets used to advance Earth system models.

In this perspective, we highlight the immense potential of omics data to advance our understanding of global biogeochemical cycles by showcasing key insights gained across diverse systems – from land to ocean to atmosphere – and addressing the remaining challenges and future opportunities. The examples are deliberately presented as standalone Boxes, one per Earth system (air, cryosphere, soil, plant, ocean), each serving as a self-contained entry point for readers working in that system.

Box 1 Glossary of omics vocabulary.

2 New insights on ecosystem dynamics and functional diversity

2.1 Global distribution of biodiversity

While around 1.5 million species have formally been catalogued to date, these estimates largely ignore microbes (both prokaryotes and eukaryotes, as well as viruses) that are estimated to encompass as many as a trillion species (Tara Ocean Foundation et al., 2022). The inability to characterize this dark matter of life is highly problematic, given microbes regulate and maintain all of Earth's biogeochemical processes including carbon (primary production in the ocean is performed almost entirely by microbes, and is comparable to all photosynthetic activity by plants on land), and nitrogen (e.g., nitrogen fixation and denitrification). Even viruses, generally not considered living organisms, are now recognized for playing major roles in nutrient recycling, e.g., carbon and iron cycles (Twining and Baines, 2013). Not only does much of the Earth's biodiversity reside in microorganisms, microbes are highly abundant (est. 10³⁰ cells on Earth) and, together, represent the second largest contributor to Earth's biomass despite their microscopic size (Bar-On et al., 2018; Bar-On and Milo, 2019). The immense amount of microbial abundance and diversity reflects billions of years of metabolic innovation and evolution to exploit nearly every resource on Earth. These key contributions of microbes as the engines of Earth's biogeochemical cycles and for assuring the survival of Earth's life are grossly neglected by most people, including by the majority of scientists, although efforts are now afoot to give a voice to this unseen majority (Falkowski and Knoll, 2007; Doumeizel and Dolan, 2024; Tara Ocean Foundation et al., 2022).

Thanks to multiple international efforts, a comprehensive picture of microbial diversity is emerging. Notable examples are the Earth Microbiome Project (Thompson et al., 2017), soil microbiome and fungal projects like SPUN (Sanchez-Tello and Corrales, 2024), and Tara Oceans (Sunagawa et al., 2020). These datasets allow us to build genomic catalogues (e.g., Nayfach et al., 2021; Paoli et al., 2022; Ma et al., 2023; Schmidt et al., 2024; Maeke et al., 2025) to characterize the patterns and processes governing microbial distributions. Gene atlases of these organisms could also be used to infer functions related to biogeochemical cycles and thus start to define who does what, how they interact with their environment, and how they impact ecosystem functioning (e.g., Guidi et al., 2016). As illustrated in the next section, such blueprints from multiple microbes provide multiple hypotheses to test the origins, current status, and future of life on Earth.

2.2 Using omics to link diversity to biogeochemical functions: a conceptual foundation

“We can say something about the community by giving a list of its species composition, but a community is poorly described by such a list alone.” (Whittaker, 1975).

Ecosystem ecology sits at the nexus of understanding the flow of energy and materials through the biosphere and other abiotic pools. Many of the biological controls on these fluxes (e.g., reaction rates, substrate affinities, and temperature sensitivities) are governed by functional traits that vary at finer genetic scales than are typically resolved in ecosystem studies, such that aggregation into coarse taxonomic units obscures functionally relevant diversity and propagates uncertainty into predictions of biogeochemical fluxes. Ecosystem ecologists have focused on the role of species variation across environmental gradients to predict ecosystem functions; for example, correlating plant species diversity with an ecosystem's measure of productivity. However, for microorganisms, we have limited knowledge of how microbial diversity translates to functional variation. For one, most microorganisms are uncultured and, therefore, remain unknown and uncharacterized (Steen et al., 2019). Second, the equivalent of the biological species concept does not apply to microorganisms, as there are ambiguities in what constitutes a microbial population, or genetic units belonging to the same species (Shapiro et al., 2016; Chase et al., 2019). Therefore, microbiome analyses typically rely on extracting all the genetic information from an environmental sample for microbial classifications, as this circumvents the need to culture and identify individual microbes. Generally, microbiome scientists cluster and collapse the immense genetic diversity within microbiomes into taxonomic units based on genetic relatedness, such as operational taxonomic units (OTUs) or metagenome assembled genomes (MAGs), which can then be compared to reference databases. However, these taxonomic groupings are defined largely by computational convenience and may not reflect biological organization, as they can represent anything from a family or species to a mosaic genome composed of multiple strains (Tyson et al., 2004; Meziti et al., 2021). Therefore, microbiome-focused ecosystem assessments are inherently forced to move away from the rich literature and theory developed for species to a broader view centered on community-wide metrics such as alpha- and beta-diversity metrics, community-level traits, or interaction metrics, e.g., with co-occurrence networks (Chase and Martiny, 2018).

This community-wide approach has several advantages, such as compiling global catalogs of microbial diversity across geographic scales that reveal biogeographic processes structuring environmental microbiomes (Fierer, 2014) and understanding potential metabolic interactions across various trophic levels encoded within the genetic makeup (Shaffer et al., 2022). Yet, these broad approaches mask intra-community functional trait variation that contributes to ecosystem functioning (Larkin and Martiny, 2017), especially in the context of environmental change (Scales et al., 2022). To illustrate this point, a bacterial OTU is assigned by clustering related genetic sequences of the conserved 16S rRNA gene region at varying thresholds. Depending on the genetic resolution used (ranging from amplicon sequence variants (ASVs) or 1 %–3 % genetic divergence), these OTUs are estimated to represent 50–150 Myr of evolutionary time, or roughly equivalent to when all modern bird species evolved from a common ancestor (Martiny et al., 2023). Given the well-established trait and physiological variation among bird species and their contributions to geographic distributions, no ornithologist would collapse all birds into a single OTU to study ecosystem dynamics. From a biogeochemical perspective, doing so would implicitly assume uniform kinetics and environmental sensitivities, an assumption rarely justified for processes such as decomposition, respiration, or primary production. As such, microbial community analyses should understand the degree of trait variation within their taxonomic designations and determine whether that functional variation is relevant to biogeochemical fluxes (Abs et al., 2023; Defrenne et al., 2021).

Given that microbiome assessments rely on genetic material, we can also utilize that genomic information to infer functional traits at multiple biological levels. All functional traits, to some degree, are phylogenetically conserved. Some, such as pH or salinity responses, are broadly conserved at higher taxonomic levels while others, such as the responses to predicted anthropogenic factors (i.e., added nitrogen or drought) (Chase et al., 2017), exhibit fine-scale genetic variation at the genus, species, or even population levels (Martiny et al., 2015). Therefore, microbiome assessments of functional consequences need to prioritize and identify their functional trait of interest and understand the system's diversity accordingly (McLaren and Callahan, 2018). For example, the globally distributed marine phototroph Prochlorococcus (Cyanobacteriota) is estimated to produce ∼20 % of the oxygen generated each year by photosynthesis on the planet and is predicted to be highly sensitive to changes in surface ocean temperatures (Flombaum et al., 2013). Prochlorococcus is composed of several ecotypes, or highly clustered strains occupying similar ecological niches but differing in key functional traits (Cohan, 2002). These ecotypes are geographically and vertically distributed in the water column through their partitioning of resources, including light, iron, and temperature, with closely related ecotypes differing in their thermal tolerance (ranging from ∼15–30 °C based on cultured isolates) and persistence across surface and subsurface waters (Johnson et al., 2006). Therefore, an understanding of how oxygen production may shift under future climate scenarios will be dependent on the individual responses of each ecotype and their functional differences (Ustick et al., 2023). These patterns aren't unique to Prochlorococcus and have been increasingly observed in abundant soil taxa driving leaf litter degradation and carbon turnover (Chase et al., 2018), glacier-fed streams (Fodelianakis et al., 2022), marine sediments (Chase et al., 2021a), and bogs (García-García et al., 2019).

Given the current limitations in relating functional and taxonomic diversity, metagenomic (DNA-based) and/or metatranscriptomic (RNA-based) assessments of microbiomes can provide direct information on the functional genes present in a community. These functional approaches provide crucial insights into how the metabolic potential of communities may be used to predict responses to environmental change (Piton et al., 2023). These data reflect genomic potential rather than realized activity unless paired with proteomics or metabolomics; however, even with such paired datasets, sparse reference databases (e.g., 90 %–99 % of environmental genes (Nayfach et al., 2016) and >95 %of metabolites are unannotated (Bouslimani et al., 2014)) can bias trait interpretations and their links to environmental variables (Osburn et al., 2024). Ultimately, scaling from traits to ecosystem processes requires understanding how variation in microbial traits – and the genetic scales at which they are conserved – links to environmental gradients and biogeochemical rates (Abs et al., 2024a, b, 2020), as ecosystem ecologists have long done for macro-organisms. While physical state variables like temperature and pH are easily monitored, they represent the environment rather than the biological response to it. Meta-omics provides the link between these abiotic drivers and the internal biological “volume knobs”, such as V_max and K_m, that determine the actual rate of biogeochemical flux.

2.3 Using omics to predict biogeochemical functions at the site scale

2.3.1 Example from a subsurface system

We define the terrestrial subsurface to begin below the depth of greatest plant root density and extend into deeper aquifers. Life in the subsurface influences the composition of gases and liquids moving through it, acting as an open biogeoreactor driven by surface-derived organic matter, nutrient inputs, and a process known as dark carbon fixation (Küsel et al., 2016). In environments lacking sunlight, dark carbon fixation refers to the ability of chemolithoautotrophic microbes to use energy from reduced compounds (e.g., sulfur, nitrogen, hydrogen) to fix inorganic carbon into organic matter, thereby generating a self-sustaining energy source. This process is particularly significant in groundwater ecosystems, where reduced inorganic electron donors fuel primary production, replacing the role of photosynthesis at the surface.

Omics approaches, such as metagenomics, are invaluable for unravelling the metabolic capacities of these microbial communities, offering insights into the dynamic and complex biogeochemical processes that drive subsurface life. Metagenomic studies have revealed a high abundance of chemolithoautotrophs in groundwater, accounting for 12 % to 47 % of the microbial community, suggesting that dark carbon fixation is central to subsurface trophic webs (Overholt et al., 2022). This process might also have global significance, as the subsurface and surface biospheres are tightly linked through the exchange of energy and matter. Overholt et al. (2022) estimated that dark carbon fixation in deep subsurface environments could contribute between 1.4 and 2.5 Tg of CO₂ per year to the atmosphere globally.

Box 2 Example from air microbiology.

2.3.2 Example from the global ocean

Meta-omic data provide powerful tools for understanding the key role of microorganisms and in carbon sequestration, biogeochemical cycles of the global ocean including data generated by the GOS (Yooseph et al., 2010), GEOTRACES (Biller et al., 2018), Malaspina (Acinas et al., 2021) and Tara Oceans (Sunagawa et al., 2020) expeditions. Among other forcings, human-induced deoxygenation of marinel water columns is altering microbial metabolisms and their contributions to global biogeochemical cycles. Oxygen levels in marine waters are declining due to global warming and increased anthropogenic nutrient loading (Doney et al., 2012). This deoxygenation is expanding oxygen-minimum zones and creating so-called dead zones in near-shore ecosystems (Breitburg et al., 2018), with profound implications for ecosystem functioning. These changes reduce habitats and resources for organisms like fish and cause significant shifts in carbon, nitrogen, and sulfur fluxes between the sea and atmosphere (Wakeham, 2020). Future deoxygenation due to climate change may expand the extent of oxygen-minimum zones with unknown consequences such as potential increase of emissions of greenhouse gazes such as nitrous oxide (Canadell et al., 2021).

Microbial metabolisms are particularly affected, as anaerobic respiration processes become dominant in oxygen-depleted environments. These metabolisms include the use of nitrogen, sulfur, carbon compounds, and trace elements as alternative energy sources. Other processes, such as anoxygenic anaerobic photosynthesis (the use of light to produce organic matter in sunlit anoxic environments) and the microbial transformation of mercury into the neurotoxin methylmercury, are also induced (Bravo and Cosio, 2020). Thanks to meta-omic data, we can now study present-day microbial ecology and biogeochemical processes associated with oxygen depletion. For example, Capo et al. (2022a) demonstrated through environmental chemistry, metagenomics, and metatranscriptomics that microbial production of methylmercury is prevalent in oxygen-deficient water layers of two basins in the Baltic Sea. Combining such contemporary meta-omic analyses with historical DNA data from sedimentary archives offers a unique opportunity to explore the links between increasing anoxia in water columns and shifts in microbial metabolisms as recently done in the Black Sea (Zhong et al., 2025).

Box 3 Example from the cryosphere.

2.3.3 Example from a terrestrial plant system

Metabolomics is also a tool to better predict the influence of climate change on ecosystem dynamics. Chemical ecologists recently included phytochemical traits as niche dimensions based on their central functions in plant responses to their environment (Müller and Junker, 2022). For instance, the so-called “predictive metabolomics” approach can be used to predict patterns of phytochemical diversity in Alpine regions independently of plant lineage (Defossez et al., 2021). Similarly, modelling was used to predict the environment of multiple plant species from the Atacama Desert using their metabolism (Dussarrat et al., 2022). Metabolomics was also used to analyse changes in soil microbial function in response to drought (Brown et al., 2021). Overall, predictive metabolomics peaks as an ideal tool to predict and anticipate losses or modifications of chemical and functional diversity in response to climate change (Dussarrat et al., 2025; Frugone-Álvarez et al., 2023; Dussarrat et al., 2022).

2.4 Using omics to predict biogeochemical functions at the global scale

2.4.1 Large-scale metagenomics

The examples in Sect. 2.2 illustrate how omics approaches can help identify the drivers of biogeochemical processes. Large-scale metagenomic analyses offer a powerful tool to extend findings from individual studies to a global scale. A variety of large-scale metagenomic datasets from diverse ecosystems are now available (Bahram et al., 2018; Sunagawa et al., 2020; Duarte, 2015; Schmidt et al., 2024; Maeke et al., 2025), allowing researchers to screen for the presence of pathways of interest at genome resolution thanks to the large numbers of MAGs that have been generated. These resources also enable us to explore global distributions, and assess community susceptibility to environmental conditions (Chaffron et al., 2021). This, in turn, helps to predict their potential impact on biogeochemical cycling at a planetary level (Fig. 2).

Figure 2 Methodological workflow for the robust screening and functional annotation of biogeochemically relevant pathways in large-scale metagenomic data. The strategy utilizes a multi-stage verification process to ensure high-specificity annotations, starting with the selection of specific marker genes and the gathering of reference sequences from known organisms and shared protein domains, such as PFAM. These reference sequences are used to phylogenetically define clades of true positives (TPs) and false positives (FPs) to build profile Hidden Markov Models (HMMs) for precise screening. Input metagenomic sequences are then screened using homology-sensitive methods, with potential hits validated by placement onto the reference phylogenetic tree to confirm their functional identity. To further handle technical uncertainty and ensure metabolic functionality, an optional final step evaluates the genomic context for operon-like organization or overall pathway completeness within the reconstructed Metagenome-Assembled Genome (MAG).

However, large-scale metagenomics presents several challenges, foremost among them being the accuracy of pathway annotations. Standard workflows rely on automated annotation pipelines, followed by manual curation to correct issues like spurious annotations. As data volume increases, however, manual curation becomes impractical. To overcome this, we propose several steps to enhance the accuracy of pathway predictions from large-scale metagenomic data. These steps include defining marker genes uniquely associated with specific pathways to serve as reliable indicators of their presence. Annotation of these genes should utilize methods sensitive to remote homology and be validated through phylogenetic approaches. Furthermore, pathway completeness should be assessed by identifying other related genes within the MAG of interest. In some cases, the operon-like organization of the genes involved in a pathway can also indicate its functionality, increasing annotation reliability (Bratlie et al., 2010). This property has been used to explore the biosynthetic potential of MAGs generated from Tara Oceans, leading to the discovery of a particularly talented group of formerly unknown bacteria able to generate multiple compounds of interest in biotechnology (Paoli et al., 2022).

Large-scale metagenomic screenings provide a comprehensive view of microbial processes across diverse ecosystems and under varying environmental conditions, thus enhancing our understanding of the phylogenetic and ecological distribution of biogeochemical cycles. Combining these with contextual data (i.e., physical, chemical and biological parameters) can help to predict the link between environmental conditions and a respective biogeochemical pathway. Furthermore, understanding the link between pathways and the environmental conditions in which they occur allows predicting the sensitivity of biogeochemical cycles in a changing environment.

In summary, we can leverage large-scale metagenomics to study the distribution of biogeochemical pathways in diverse environments and make predictions of how their distribution might change under changing environmental conditions. The results from such analyses provide the foundation for new hypotheses and experimental research (e.g., isolation of organisms or studying pathway activity in different conditions).

2.4.2 Combining meta-omics to infer large scale biogeochemistry

In most Land Surface Models (LSMs), vegetation and soil attributes are often prescribed in parameter lists and used to describe how biogeochemical processes vary over the land surface. For example, the representation of vegetation is distilled in a variety of plant functional types (PFTs), with each representing a prescribed range of anatomical and physiological parameter values such as leaf mass area, carboxylation rates, or leaf respiration (Duckworth et al., 2000; Woodward and Cramer, 1996). These PFTs can co-vary with certain soil chemical and physical characteristics that are often prescribed, assimilated and/or predicted after some model initialisation and predetermined spin up procedure that drives the vegetation, soil, and atmosphere into some quasi-steady-state to explore feedbacks on biogeochemical cycles (Friend et al., 2014; Matthaeus et al., 2021; Taylor et al., 2011).

A growing number of databases and maps detailing key soil characteristics – such as soil bulk density, pH, carbon (C), nitrogen (N) and phosphorus (P) content – combined with climate forcing, are now openly available. These resources can support LSM numerical simulations, machine learning models, or hybrid approaches, to predict how soil properties, such as microbial C, N and P biomass, vary across the land surface. These soil characteristics are increasingly being linked to variations in soil microbial community structure and diversity. For example, Delgado-Baquerizo et al. (2016) found that certain microbial groups thrive in high-pH soils, itself shaped by vegetation and climate. Similarly, fungal communities can be mapped to how soil mycorrhizal fungal guilds will vary across the land surface, with implications for soil C storage and nutrient acquisition by vegetation (Kivlin, 2020).

Figure 3 Conceptual figure of a bottom-up system approach using meta-omics. The integration and combination of omic datasets with key vegetation and soil attributes through machine learning models present a great opportunity to enhance our understanding of the genes, pathways and mechanisms underlying biogeochemical cycling and functional activity of soil microbial communities and associated processes such as soil gas exchange for their representation at larger scales. Created with BioRender (https://BioRender.com/9p03o4f, last access: 29 July 2026).

More recently, these databases have been extended to include metagenomic data to show how genes linked to C storage vary across the global land surface (Piton et al., 2023) (Table 1). Linking metagenomic data with measurements of soil GHG fluxes or profiles of soil volatilomes could provide mechanistic insight into the metabolic characteristics of soil communities that control GHG and VOC fluxes with the atmosphere. These, in turn, could offer predictive tools for understanding how GHG and VOC fluxes might change with climate or disturbance (Fig. 3). Likewise, other omic datasets, such as soil proteomes or metabolomes, may provide additional insights into the realized and functional activity of soil microbial communities. For example, examining how soil GHG fluxes and gridded soil characteristics vary with soil metabolomes, proteomes, and community structure could pave the way for predicting soil C storage and GHG variability across large scales and over time, even from discrete soil samples.

Beyond direct parameterization, emerging machine-learning (ML) and AI-based models provide a powerful bridge between microbial community structure and ecosystem functions. For example, ML frameworks have been successfully used to map bacterial communities to physicochemical variables and soil quality indicators, allowing for the prediction of ecosystem-level states from sequence data (Hermans et al., 2020). These predictive models can be further expanded to forecast environmental shifts, such as changes in soil pH in response to global warming, by training models on large-scale longitudinal datasets (Feng et al., 2024). Complementary to ML, genome-informed approaches allow for the inference of specific environmental preferences, such as microbial pH optima, directly from genomic traits (Ramoneda et al., 2023). By linking these inferred traits to functional outcomes, researchers can move beyond taxonomic lists to mechanistic representations of how microbial “niche breadths” influence biogeochemical stability.

Table 1 Databases and resources for linking omics data to ecosystem processes. The table provides an illustrative, non-exhaustive overview of widely used resources across ecosystems, including data type, ecosystem coverage, strengths and limitations, access standards, and references where applicable.

2.4.3 Integrating omics data into biogeochemical models

Traditional biogeochemical models often apply simplified representation of microbial processes, relying on parameters derived from cultivated microorganisms, which may not accurately capture the complexity and diversity of natural ecosystems (Steen et al., 2019). The increasing availability of metagenomics data presents an opportunity to integrate those microbial traits from genomes into mechanistic ecosystem models, potentially enhancing their predictive capabilities (Woodcroft et al., 2018; Bahram et al., 2018). By inferring microbial traits from genomes obtained in-situ, the metabolic potential of the microbial communities would be better represented, thereby providing more accurate and robust prediction of ecosystem functioning.

Incorporating genomic information into biogeochemical models requires a structured approach. Here, we review two groups of models: soil biogeochemical models, and full ecosystem-level models, each with distinct capabilities and approaches leveraging omics data. Soil biogeochemical models are designed to simulate detailed soil processes, and are typically driven by diagnostic inputs, such as prescribed litter, soil moisture, and temperature data. Some of these models have explicit representations of microbial processes, including the Microbial-Enzyme Decomposition (MEND) Model, MIcrobial-MIneral Carbon Stabilization (MIMICS), and CORPSE; see Chandel et al. (2023) for a full review. For example, MEND has been constrained and validated using GeoChip-based gene abundance measurements, which provides a basic platform for testing hypotheses about microbially-mediated biogeochemical processes (Wang et al., 2022a; Gao et al., 2020). In contrast to soil biogeochemical models, full ecosystem models simulate the coupled dynamics of water, energy, plants, and soil processes driven by climate forcing. Such coupled models are valuable for projection or spatial extrapolation because many of the emergent properties of primary scientific and societal concern, such as net ecosystem exchanges of CO2 and CH4, are strongly regulated by interactions between these interconnected systems. One example of such a model that integrates microbial genomics data is the genome-to-ecosystem (G2E) framework proposed in Li et al. (2025) (Box 4).

Box 4 A worked example: genome-to-ecosystem (G2E) framework.

Figure 4 Genome-to-ecosystem (G2E) conceptual framework integrating genomic information, site characterization, and ecosys. Reproduced from Li et al. (2025).

2.5 Using omics to reconstruct past and forecast future environments

2.5.1 Past

Omics methodologies in paleo-ecological reconstructions have advanced from mere novelty to producing new ways of understanding ecosystem functions. Compared with traditional palynology techniques, such as pollen, fungal spores, and bacteria microfossils, metagenomics for instance captures changing trophic relationships among primary producers, consumers, and decomposers albeit at limited spatial scales (Jia et al., 2022). Still, omics and conventional proxies can be successfully combined despite the variations in spatial scale of processes and taxonomic resolution. For example, omics have been utilized to differentiate past ecosystem drivers and expand assessments of vegetation changes in rangeland ecosystems (Box 5). However, combining traditional and omics evidence in palaeo-ecology is still in its infancy because there are few epistemically independent models (Jones and Bösl, 2021; Edwards, 2020).

Molecular paleomicrobiology is an emerging field that leverages ancient and/or historical DNA (aDNA) technologies to uncover how microbial genomes and communities have evolved over time in response to environmental changes (Grasso et al., 2024). By studying ancient microbial remains preserved in dental calculus, paleofeces, archaeological sediments, and even environmental samples like permafrost, aquatic sediments and ice cores, scientists gain insights into historical human-microbe interactions and environmental adaptations (Warinner et al., 2014, 2017; Warinner, 2022; Capo et al., 2022b). Remarkably well-preserved microbial aDNA from stable environments, such as permafrost, allows us to track the evolutionary history of microbial taxa and understand ecosystem responses to past climate shifts. For instance, DNA from two-million-year-old Greenland permafrost has revealed how plant and animal communities from the Middle Pleistocene adapted to environmental changes. However, the complex mixture of DNA from ancient and more recent microbial populations still complicates efforts to disentangle the signals from pioneering microbes and those deposited at a later time (Fernandez-Guerra et al., 2023; Kjær et al., 2022).

While post-mortem DNA degradation complicates the analysis, predictable damage patterns, such as fragmentation and specific base modifications, offer clues to the age of the samples (Briggs et al., 2007). This information is crucial as we work to reconstruct microbial genomes from diverse environments, which might contain amounts of aDNA that may reflect past ecosystems. Recently, DNA could be even extracted from low-biomass containing carbonate rocks with distinct DNA fragmentation patterns allowing us to separate ancient from modern communities (Wegner et al., 2023). Thus, these rock formations might act as potential archives of ancient microbial records. However, MAGs have to be reconstructed from these challenging environments to deepen our understanding of subsurface microbial life and its long-term changes, ultimately refining estimates of microbial populations in the Earth's crust.

Long-term records of aquatic biota are still scarce as monitoring using DNA approaches started about two decades ago. Fortunately, the extraction and sequencing of environmental DNA preserved in sedimentary archives has shown its potential to reconstruct past environmental changes and their impact on aquatic microbial diversity (Nguyen et al., 2023; Capo and Barouillet, 2023). In a recent review, Barouillet et al. (2022) highlighted how molecular paleoecology research augments knowledge about the effects of anthropogenic stressors on past aquatic biota. On a longer time scale, Armbrecht et al. (2022) show that photosynthetic genes dated back to 540 000 years ago can be detected from Southern Ocean sedimentary archives. Overall, past DNA data is foreseen to be a valuable proxy for projecting the impacts of near-future environmental changes on microbial diversity across both aquatic and terrestrial systems.

Box 5 Sedimentary ancient DNA in Rangelands.

Figure 5 A graphical representation of trophic energy flow in African savanna rangelands featuring producers, consumers, and decomposers. The gray arrow indicates the flow of energy while the colour-coded trophic level boxes represent trophic biomass (not to scale) across the chain. Vegetation dynamics of this system can be reconstructed from fossil pollen, which can be related to herbivore grazing pressure and nutrient cycling via coprophilous fungal spores. Metagenomics, in contrast, is useful for resolving protein and genetic sequences, simultaneously tracing multiple trophic interactions. However, since preservation and strengths of trophic relationships may vary within ecosystems, combining palynology and omics produces better results (e.g., Tabares et al., 2020).

2.5.2 Future

Experimental approaches to studying microbial evolution and its role in biogeochemistry

Geological and paleoclimate records provide environmental analogs to how ecosystems may respond to future climate scenarios. However, such studies are typically limited in their scope by generalizing at the ecosystem-level without insights into the processes and mechanisms driving individual species responses. One such response is local adaptation to changing environmental conditions, where species may disperse to new favourable conditions and expand their geographic ranges or evolve adaptive traits over evolutionary time through gene flow with neighbouring populations and/or emergence of advantageous de novo mutations. Given the long timescales to observe adaptive traits, reciprocal transplant experiments have provided direct insights into the ecological and evolutionary responses to predicted climate conditions. For example, plant communities, given their importance in terrestrial ecosystems, have long been studied demonstrating plant species are locally adapted to current environmental conditions. Such studies highlight differential survival and fecundity (i.e., flowering time) to changing conditions, highlighting phenological traits that may contribute to adaptive differentiation and possible evolutionary responses (Ågren et al., 2012). In other words, past evolutionary divergence caused by local adaptation can restrict and influence contemporary ecological patterns and processes (Urban et al., 2020). Extending these approaches to explicitly test how microbial ecological and evolutionary responses alter biogeochemical rates represent a critical next step for predicting ecosystem responses to future climate change (Box 6).

Numerous microbial studies have demonstrated, through reciprocal transplant experiments or environmental manipulations (e.g., drought or temperature), that microbial communities can rapidly respond through the ecological process such as species sorting, defined by the differential survival and reproduction of locally adapted taxa (e.g., DeAngelis et al., 2015; Finks et al., 2021). These taxonomic shifts can have direct implications for ecosystem function, including changes in decomposition rates (Glassman et al., 2018) and reductions in carbon use efficiency under long-term soil warming (Li et al., 2019). In addition, dispersal can play a significant role in microbiome recovery and successional dynamics following large-scale disturbances (i.e., wildfires; Barbour et al., 2023), which are expected to increase in frequency and intensity under future climate scenarios.

However, microbes are distinct in that their generation times, and thus capacity for adaptation, operate on much faster timelines. Combined with large population sizes and their capacity to exchange DNA within and across lineages, microbes are exposed to a vast reservoir of genetic diversity within microbial communities that may contribute to rapid adaptive responses to environmental change. Certainly, lab-based experimental evolutionary studies that manipulate a selective pressure (e.g., heat) have repeatedly demonstrated that microbes can rapidly evolve and adapt over weeks to months (Tenaillon et al., 2012; Kent et al., 2018), although it's unclear how these artificial laboratory conditions translate to natural populations. In contrast to species sorting, which alters the relative abundances of existing trait variants, evolutionary responses modify trait values themselves, potentially changing microbial reaction norms and longer-term sensitivities to environmental change. Recently, natural experiments using a reciprocal transplant approach across a regional climate gradient demonstrated that microbes, as observed in lab-based experiments, can rapidly adapt through de novo mutations in response to predicted climate scenarios (Chase et al., 2021b). However, such adaptive responses under natural conditions are likely operating under reduced selective pressures on monthly and annual timescales and, thus, may have a reduced role compared to rapid ecological responses (Chase et al., 2021b). Nonetheless, a shift in the microbiome under changing climate conditions is dependent on the continuum of ecological and evolutionary processes that operate at the same time. It is therefore the consideration of the relative influence of ecological and evolutionary processes – and the timescales over which they interact – that will be necessary to predict how microbial communities regulate biogeochemical rates and ecosystem functioning under environmental change (Martiny et al., 2023; Abs et al., 2024b, c).

Predicting long-term ecosystem responses using omics-parameterized models

Using microbial kinetic traits inferred from multi-year omics data, ecosys can be used to predict future ecosystem dynamics. Specifically, ecosys was employed to predict ecosystem dynamics in an Arctic wetland under the RCP8.5 scenario (RCP: Representative Concentration Pathway) through the year 2100 (Li et al., 2025). That ecosys simulation incorporated genome-inferred microbial traits (i.e., community-aggregated traits weighted by genome relative abundance) and baseline microbial traits. The model results show reasonable seasonal cycles of methane and net ecosystem exchange (NEE) during the late 21st century, and have about 18 % and 21 % difference in annual methane emissions and NEE, respectively. The differences were mediated by shifts in the overall carbon cycle, largely due to microbial activity, belowground nutrient transformations, and vegetation dynamics (Bouskill et al., 2020). We highlight some caveats to these results. First, the microbial traits inferred from omics that we applied are constants based on current life-history traits (e.g., growth rate, resources utilization, and stress tolerance); however, those traits may adapt to changing environmental conditions (Sokol et al., 2022). Work is underway to develop models of how adaptation may occur, and to integrate those approaches with ecosystem models (Abs et al., 2024b, 2025; Abs and Ferrière, 2020; Schwarz et al., 2025). Second, the genome-inferred microbial traits we determined span ranges within each microbial functional group, reflecting the multiple genomes that perform the same metabolic functions (Li et al., 2025). However, our application of these traits described here considered only a single value for each trait in each microbial group. Ongoing work in ecosys will allow for trait diversity within microbial functional groups, thereby allowing prediction of dynamic community assembly under dynamic and competitive environmental conditions.

Box 6 Predicting future ocean biogeochemistry.

Figure 6 Projected effects of diversity changes within plankton communities on marine ecosystems, fisheries, and biogeochemical cycles (adapted from Ibarbalz et al., 2019). The panel in (a) summarizes the biodiversity anomalies projected by end-century from Tara Oceans data for a range of different plankton groups (indicated by the different colours), and emphasizes that these projections indicate major changes at mid- to high latitudes (grey bars). The global maps in panel (b) indicate estimates of carbon export in different oceanic regions (left), fish landings (centre), and the locations of Marine Protected Areas (right), all of which are likely to be impacted significantly by the projected biodiversity anomalies shown in a (grey bars). Figure adapted from Ibarbalz et al. (2019), Tara Ocean Foundation et al. (2022).

3 Challenges and perspectives

Long-term Earth system monitoring programs, along with preserved sample archives, present a valuable opportunity to enhance Earth System Models (ESMs) by providing insights into climate-driven ecosystem transformations. The recent surge in omics-based technologies has catalysed untargeted measurements across diverse environments, as highlighted in this paper with examples from deep-sea sediments, plant canopies, the atmosphere, cryospheric regions and marine ecosystems. However, the proposed implementation of omics driven data into modelling efforts faces challenges outlined in the following.

3.1 Data integration and interpretation

Integration of omics data into ESMs is complex due to the sheer diversity, scale, and volume of data these approaches generate. Multi-omics datasets capture different biological layers, often involving vast amounts of heterogeneous data. This heterogeneity arises because each of these layers represents different types of biological information, from genetic sequences to metabolite profiles and often vary in data structure, format, and the presence of missing information due to limitations inherent in omics technologies (Ramirez Flores et al., 2023).

Connecting these diverse datasets to critical ESM response variables – such as GHG emissions, soil carbon stocks, or other processes relevant to predicting human-induced climate change – is challenging. For instance, microbial-driven methane production is well characterized: its defined enzymatic pathways and environmental controls might enable more straightforward integration of omics data. In contrast, turnover of soil organic carbon is driven by a highly diverse array of microbes operating under variable conditions. This complexity might make it much harder to translate omics-based measurements into accurate process rates for ESM incorporation (He et al., 2024a).

Recent advances in deducing microbial traits from metagenomic data show great promise for integrating omics information into biogeochemical models by linking these data with kinetic parameters (Karaoz and Brodie, 2022). By identifying functional traits from metagenomes, researchers can begin to quantify the roles of various microbes in processes like nutrient cycling or carbon turnover, which holds the potential to enhance the predictive power of ESMs. Extending this approach to other types of omics data could lead to more comprehensive ESMs that incorporate detailed biotransformation kinetics, ultimately resulting in more accurate climate-related predictions. The trade-off between the speed of direct observation and the depth of omics sequencing is a matter of descriptive vs. predictive capacity. Direct measurements are essential for tracking current budgets, but physical state changes often lag behind the microbial shifts that drive them. In contrast, characterizing functional potential allows models to account for the “legacy effects” and “lagged responses” of communities, ensuring that projections remain accurate even as historical environmental correlations break down.

To navigate these complexities and ensure data reliability, researchers employ a standardized toolkit and validation strategy:

• Quality Control & Assembly. Raw sequences are processed to remove noise (QC) before being assembled into contiguous sequences (contigs).

• Genome Reconstruction & Binning. Contigs are grouped into Metagenome-Assembled Genomes (MAGs) and evaluated for completeness and contamination using tools like CheckM.

• Trait Inference. Bioinformatic pipelines like microTrait or DRAM are used to distill genomic potential into fitness traits and metabolic parameters (V_max, K_m).

• Validation & Uncertainty. Model results are confronted with in-situ flux measurements (e.g., from NEON or ICOS) to quantify the accuracy of microbial process representations.

• aDNA Authenticity. In molecular paleomicrobiology, specific damage patterns (e.g., fragmentation and base modifications) are used to discriminate ancient genetic signals from modern contamination.

3.2 Data accessibility and standardization

Ensuring accessibility to omics data and establishing robust community standards constitute essential steps to advance the integration of omics data into ESMs. Initiatives like the National Microbiome Data Collaborative (NMDC) (NMDC, 2022) and Integrated Microbial Genomes & Microbiomes (IMG/M) (Chen et al., 2024) created frameworks that promote data sharing and enforce metadata standardization. These frameworks provide structured approaches for documenting and sharing omics data, ensuring that researchers follow uniform guidelines when collecting, annotating, and storing data. The interoperability they promote relies on community-agreed reporting standards and controlled vocabularies. The most widely adopted is the Minimum Information about any (x) Sequence (MIxS) standard, developed by the Genomic Standards Consortium, which specifies the minimum contextual metadata (habitat, geolocation, and sequencing methodology) required to describe genomes, metagenomes, and marker gene sequences through its associated checklists (e.g., MIGS for genomes, MIMS for metagenomes) (Yilmaz et al., 2011). MIxS is frequently paired with the Environment Ontology (ENVO), which provides a controlled vocabulary for the consistent annotation of biomes, environmental features, and materials such as soil, water, and air (Buttigieg et al., 2013). Other domains rely on complementary standards, such as the ISA framework (ISA-Tab) for metabolomics and the ISO 19115 and Climate and Forecast (CF) conventions for geospatial and atmospheric data, reflecting the heterogeneity of standards that a unified cross-ecosystem framework would need to reconcile. However, the adoption of these standards across the global scientific community has been limited so far. Such a globally implemented and executed framework would help address questions on large scales, as it would enable the aggregation and cross-referencing of omics data from diverse ecosystems worldwide. Additionally, standardized data formats would also facilitate data access to emerging AI-driven tools which hold the potential to play a key role in identifying yet unseen barriers in improvements of ESMs and might also help to overcome those.

3.3 Emerging technologies

Technological advances, particularly in ancient/historical DNA (aDNA) analysis, have opened new pathways for studying long-term ecological responses to environmental changes. aDNA analysis enables scientists to retrieve genetic information from organisms preserved in ancient samples, such as ice cores (Zhong et al., 2021) and up to 2 million year old sediment layers (Nguyen et al., 2023; Kjær et al., 2022; Fernandez-Guerra et al., 2023). By integrating biological data with chemical signatures found in these preserved samples, we can reconstruct ecosystem dynamics over extended historical periods, offering valuable insights into how biomes have responded to past climate variations. Such and other omics data will become increasingly available now as ongoing advancements in omics technology have gradually reduced the amount of analyte required for analysis, making it feasible to use small or degraded samples from ancient archives.

Incorporating these ancient biological signatures into ESMs has the potential to enhance model accuracy significantly. By cross-validating current ESM data with insights from historical records, scientists can better understand the relationships between abiotic factors and biotic responses over time. This integrative approach could improve the models' capacity to predict ecological responses to future climate changes by anchoring them in real historical data, providing a long-term perspective that modern data alone cannot offer.

3.4 The necessity of interdisciplinary collaboration

Unsurprisingly, integrating omics data into ESMs demands collaboration across disciplines like ecology, geochemistry, bioinformatics, modelling, and data science. Each field brings crucial expertise: ecologists and geochemists link biological and chemical processes, bioinformaticians manage complex omics data, modelers translate findings into model parameters, and data scientists develop frameworks to address challenges in data integration and accessibility. Environmental scientists play a key role in identifying biochemical pathways and traits essential for model parameterization, thereby improving the accuracy of ESMs in simulating ecosystem functions.

Collaborative efforts like these would produce enhanced ESMs that incorporate omics data, rooted in comprehensive, relevant datasets and overcoming the barriers of data integration and accessibility. We believe that the EGU community offers an ideal platform for such interdisciplinary initiatives. Leveraging existing long-term ecological monitoring projects through collaborative projects generating omics data aimed for model integration, would be a logical first step in these efforts, and we encourage all stakeholders to use this opportunity and actively participate.